Method and mechanism to reduce handwriting recognizer errors using multiple decision trees

• US 6,973,215 B1
• Filed: 03/20/2000
• Issued: 12/06/2005
• Est. Priority Date: 06/06/1997
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Hullender, Gregory N., Haluptzok, Patrick M., Bennett, John R.
• Primary Examiner(s): Bali, Vikkram
• Application Number:
  US09/528,889
• Time in Patent Office: 2,087 Days
• Field of Search: 382/187, 382/188, 382/189, 382/190, 382/186, 382/185, 382/226, 382/227, 382/157, 345/176
• US Class Current: 382/187
• CPC Class Codes: G06F 18/254 of classification results, e.g. of results related to same input data
• CPC Class Codes: G06V 30/347 Sampling; Contour coding; Stroke extraction
• CPC Class Codes: G06V 30/373 using a special pattern or subpattern alphabet

Abstract

An improved method and mechanism for recognizing chirographs (handwritten characters) input into a computer system. A primary recognizer is provided for converting chirographs to code points, and secondary recognizers such as binary CART trees are developed and trained to differentiate chirographs which produce certain code points at the primary recognizer. Each such secondary recognizer is associated with each selected code point. When a chirograph is received, the chirograph is provided to the primary recognizer whereby a code point corresponding thereto is received. If the code point corresponds to one of the secondary recognizers, the chirograph is passed to the secondary recognizer, and a code point is returned from the secondary recognizer. If not, the code point provided by the primary recognizer is returned. The invention sets forth an automated process for training the CART trees and for optimizing the recognition mechanism by discarding CART trees which do not improve on the recognition accuracy of the primary recognizer.

Claims

1. A method of recognizing chirographs input into a computer system, comprising:

providing a primary recognizer for converting chirographs to shape indexes, the primary recognizer providing output including a shape index when a chirograph is input thereto;

providing a plurality of secondary recognizers to convert chirographs into code points, and associating the secondary recognizers with at least some of the shape indexes;

receiving a chirograph;

providing the chirograph to the primary recognizer and receiving a shape index therefrom;

determining by a separate look-up process whether one of the secondary recognizers is associated with the shape index, and if so, selecting that secondary recognizer as a selected secondary recognizer and passing the chirograph to the selected secondary recognizer, the secondary recognizer returning a code point; and

training the secondary recognizers by providing a first training set comprising a plurality of chirographs and actual code points for each chirograph, wherein training the secondary recognizers further comprises determining a plurality of distinguishing features of chirographs based on predetermined criteria corresponding to questions, and wherein training the secondary recognizers further comprises determining a question ordering by measuring the quality of each question.

2. The method of claim 1 wherein the shape index comprises a code point.

3. The method of claim 1 wherein at least one of the secondary recognizers comprises a CART tree.